An Integration Method of Expert Experience and ADT Data Based on Uncertain Cross-Entropy

摘要

Accelerated degradation testing (ADT) is widely applied to evaluate product reliability and lifetime before releasing products to the market. The degradation data from ADTs are usually integrated with expert experience in the development phase for high accurate reliability assessments. However, in practical applications, expert experience may only provide subjective knowledge, and the degradation data are usually scarce due to the limited test resources in conducting ADTs. So epistemic uncertainties are embedded in both expert experience and ADTs. Therefore, the integration of these two information sources with epistemic uncertainties for high accurate reliability assessments is an urgent issue to be solved. To address such issue, a novel uncertain information integration method of expert experience and ADT data based on uncertain cross-entropy is presented. First, the uncertainty theory is introduced to obtain the lifetime characteristic distribution from expert opinions and to model the degradation process from ADT data. Then, the reliability information of these two sources is integrated for reliability assessments based on uncertain cross-entropy. A case study of the integration of expert experience and stress relaxation ADT data for electrical connector is applied to verify the effectiveness of the proposed method.